Re: [PATCH 4/4] x86/xen: drop USERGS_SYSRET64 paravirt call

[Andy Lutomirski]

On Mon, Nov 16, 2020 at 7:23 AM Juergen Gross <jgross@xxxxxxxx> wrote:
>
> USERGS_SYSRET64 is used to return from a syscall via sysret, but
> a Xen PV guest will nevertheless use the iret hypercall, as there
> is no sysret PV hypercall defined.
>
> So instead of testing all the prerequisites for doing a sysret and
> then mangling the stack for Xen PV again for doing an iret just use
> the iret exit from the beginning.
>
> This can easily be done via an ALTERNATIVE like it is done for the
> sysenter compat case already.
>
> While at it remove to stale sysret32 remnants.
>
> Signed-off-by: Juergen Gross <jgross@xxxxxxxx>

Acked-by: Andy Lutomirski <luto@xxxxxxxxxx>

FWIW, you've lost the VGCF_in_syscall optimization.  Let me see if I
can give it back to you better.

> ---
>  arch/x86/entry/entry_64.S             | 22 +++++++++-------------
>  arch/x86/include/asm/irqflags.h       |  6 ------
>  arch/x86/include/asm/paravirt.h       |  5 -----
>  arch/x86/include/asm/paravirt_types.h |  8 --------
>  arch/x86/kernel/asm-offsets_64.c      |  2 --
>  arch/x86/kernel/paravirt.c            |  5 +----
>  arch/x86/kernel/paravirt_patch.c      |  4 ----
>  arch/x86/xen/enlighten_pv.c           |  1 -
>  arch/x86/xen/xen-asm.S                | 20 --------------------
>  arch/x86/xen/xen-ops.h                |  2 --
>  10 files changed, 10 insertions(+), 65 deletions(-)
>
> diff --git a/arch/x86/entry/entry_64.S b/arch/x86/entry/entry_64.S
> index a876204a73e0..df865eebd3d7 100644
> --- a/arch/x86/entry/entry_64.S
> +++ b/arch/x86/entry/entry_64.S
> @@ -46,14 +46,6 @@
>  .code64
>  .section .entry.text, "ax"
>
> -#ifdef CONFIG_PARAVIRT_XXL
> -SYM_CODE_START(native_usergs_sysret64)
> -       UNWIND_HINT_EMPTY
> -       swapgs
> -       sysretq
> -SYM_CODE_END(native_usergs_sysret64)
> -#endif /* CONFIG_PARAVIRT_XXL */
> -
>  /*
>   * 64-bit SYSCALL instruction entry. Up to 6 arguments in registers.
>   *
> @@ -123,12 +115,15 @@ SYM_INNER_LABEL(entry_SYSCALL_64_after_hwframe, SYM_L_GLOBAL)
>          * Try to use SYSRET instead of IRET if we're returning to
>          * a completely clean 64-bit userspace context.  If we're not,
>          * go to the slow exit path.
> +        * In the Xen PV case we must use iret anyway.
>          */
> -       movq    RCX(%rsp), %rcx
> -       movq    RIP(%rsp), %r11
>
> -       cmpq    %rcx, %r11      /* SYSRET requires RCX == RIP */
> -       jne     swapgs_restore_regs_and_return_to_usermode
> +       ALTERNATIVE __stringify( \
> +               movq    RCX(%rsp), %rcx; \
> +               movq    RIP(%rsp), %r11; \
> +               cmpq    %rcx, %r11;     /* SYSRET requires RCX == RIP */ \
> +               jne     swapgs_restore_regs_and_return_to_usermode), \
> +       "jmp    swapgs_restore_regs_and_return_to_usermode", X86_FEATURE_XENPV

I'm not in love with this save-a-few-bytes stringify, but I can live with it.

--Andy